performance-budgeting

Performance Budgeting

Purpose

Set and manage frame, load, memory, streaming, or bandwidth budgets before performance debt becomes structural.

Use When

• target hardware is defined
• performance problems are emerging
• content scale is increasing

Inputs

• target platforms
• feature set
• technical architecture
• content assumptions

Process

1. set budgets by platform and major runtime mode
2. allocate budget ownership by system or content class
3. choose representative benchmark scenarios
4. review regressions at milestone checkpoints
5. link descopes or optimization work to measured overages

Outputs

• budget sheet
• owner-by-budget map
• benchmark scenarios
• regression review cadence

Quality Bar

• budgets are set per system (frame time, memory, draw calls, asset sizes) against the weakest target device
• every budget has an owner and a repeatable measurement method
• regressions are caught by measurement at integration time, not discovered in QA
• budget exceptions are negotiated and recorded, not silently absorbed

Common Failure Modes

• budgets defined as totals with no per-system allocation, so no one owns overruns
• profiling only on developer hardware, missing the real target floor
• performance treated as a polish-phase task instead of a standing budget
• one-off optimizations that decay because no measurement guards them

Related Agents

• performance-reviewer
• producer
• technical-artist

Related Commands

• perf-budget
• verify
• release-check

Notes

• Keep this skill aligned with the relevant rules layer and current project documentation.
• If engine-specific constraints materially change the workflow, hand off to the matching engine skill or engine-specific reviewer.